geminispace.info

commit 04785c1de606b8e05dcfc8b5c5854b0cea3afb3b
parent b2a0e6afea9f710e6d1c0a61177a337384124223
Author: Natalie Pendragon <natpen@natpen.net>
Date:   Sun,  7 Jun 2020 08:32:16 -0400

[crawl] Ignore some more things

Diffstat:
M
gus/crawl.py
 | 
21
+++++++++++++++++++++

1 file changed, 21 insertions(+), 0 deletions(-)
diff --git a/gus/crawl.py b/gus/crawl.py
@@ -103,10 +103,31 @@ EXCLUDED_URL_PREFIXES = [
 
     # Marmaladefoo calculator
     "gemini://gemini.marmaladefoo.com/cgi-bin/calc.cgi?",
+
+    # Individual weather pages
+    "gemini://acidic.website/cgi-bin/weather.tcl?",
+
+    # Alex Schroeder's problematic stuff
+    "gemini://alexschroeder.ch/image_external",
+    "gemini://alexschroeder.ch/comments_on",
+    "gemini://alexschroeder.ch/http",
+    "gemini://alexschroeder.ch/https",
+    "gemini://alexschroeder.ch/tag/",
+    "gemini://alexschroeder.ch/raw/",
+    "gemini://alexschroeder.ch/map/",
+    "gemini://alexschroeder.ch/do/rc",
+    "gemini://alexschroeder.ch/do/rss",
+    "gemini://alexschroeder.ch/do/new",
+    "gemini://alexschroeder.ch/do/more",
+    "gemini://alexschroeder.ch/do/tags",
+    "gemini://alexschroeder.ch/do/match",
+    "gemini://alexschroeder.ch/do/search",
 ]
 
 EXCLUDED_URL_PATHS = [
     "atom.xml",
+    "favicon.ico",
+    "favicon.txt",
     "robots.txt",
     "rss.txt",
     "rss.xml",